Age at Arrival, Parents and Neighborhoods: Understanding...
Transcript of Age at Arrival, Parents and Neighborhoods: Understanding...
Age at Arrival, Parents and Neighborhoods:Understanding the Educational Attainment of
Immigrants’ Children∗
Laura Ansala Ulla Hamalainen Matti Sarvimaki
May 16, 2019
Abstract
This paper documents how children of immigrants cope in early adulthood inFinland. We first show that, on average, they have dramatically lower educationalattainment than children of natives do. This difference can be attributed to age atarrival, parental income and neighborhood quality. In fact, Finnish-born childrenof immigrants obtain more education than children of natives growing up in thesame zip codes and in families with similar parental income and family structure.Sibling comparisons suggest that age at arrival has a causal impact on educationalattainment. The results are very similar for the likelihood of being idle and havingbeen convicted of a crime, but are starkly different for the likelihood of having beenreimbursed for psychotropic medication.
Keywords: children of immigrants, second-generation immigrants, education, age atarrival. JEL Classification: I14, I21, J15
∗We thank Olof Aslund, Kari Hamalainen, Briitta van Staalduinen, and seminar participants at theFDPE Workshop on Public Economics and Labour Economics for helpful discussions and comments, andthank the Strategic Research Council at the Academy of Finland (grant 293445) for financial support.An earlier version of this paper was circulated as ”Slipping through the Cracks of a Welfare State? Chil-dren of Immigrants in Finland” (CReAM DP 06/16). Affiliations and emails. Ansala: City of Helsinki,[email protected]; Hamalainen: Ministry of Finance, [email protected]; Sarvimaki: Aalto Univer-sity School of Business and VATT Institute for Economic Research, [email protected]
1 Introduction
In continental Europe, children of immigrants tend to have lower educational attainment
and achievement than do children of natives, particularly among immigrant groups in
which parents struggle in the labor market (Algan et al. 2010; Dustmann et al. 2012;
Schnepf 2007). This stylized fact has given rise to concerns about whether the offspring
of immigrants will integrate into host countries’ societies even in the long term. Indeed,
claims about the perceived lack of integration and appropriate policy responses are among
the most important, and controversial, topics of many countries’ public debates.
Designing efficient policies requires informed hypotheses about the reasons for why
so many children of immigrants fall behind. One possibility is that the educational gaps
simply reflect the fact that immigrants’ children often grow up in low-income families and
in deprived neighborhoods. In this case, policy makers might be best advised to focus on
general policies attempting to help disadvantaged children (regardless of their parents’
immigrant status). Policies helping adult immigrants in the labor market might also have
spillover effects for their offspring. Alternatively, at least a subset of immigrants’ children
might require special attention. In this case, the first step of meaningful policy formation
is to understand which groups are the most vulnerable.
This paper adds to the debate by documenting how the offspring of immigrants cope
in early adulthood in Finland. Our primary purpose is to establish a set of facts that
are directly relevant for policy and that can be used as a basis for future research. We
address two broad questions. First, we ask how large the immigrant-native gaps in
educational attainment are, how much heterogeneity there is across different immigrant
groups, and how much of these gaps can be attributed to differences in other background
characteristics, neighborhood quality and age at migration. Second, we examine whether
the differences in education also extend to other domains.
We start by documenting large differences in educational attainment between children
of immigrants and children of natives. For example, less than half of immigrants’ children
have completed secondary education by age 20 (in comparison to four fifths among the
children of natives). Part of this gap is due to children of immigrants progressing more
slowly through their education. Nevertheless, even for immigrant children at age 23, we
find a 21 percentage point gap in the likelihood of having graduated from high school or
vocational education, and a 17 percentage point gap in the likelihood of having attended
1
college.
The unconditional differences in educational attainment contain many layers of hetero-
geneity. We start unpacking them by showing that Finnish-born children of immigrants
obtain much more education than do those immigrating with their parents. We also
document large differences across source areas. Children of immigrants from Somalia,
Turkey and Iraq tend to have particularly low educational attainment.
In the last step of our baseline analysis, we compare children of immigrants with
native children who grew up in the same zip codes and in similar families in terms
of household structure, parental income and parental employment. We find that, on
average, second-generation immigrants are as likely to hold an upper secondary degree
and are more likely to attend college than are members of this comparison group. The
only subgroup for which we find a statistically significant negative difference is that of
children of immigrants from Somalia.
These observations suggest that it might be necessary to design policies tailored for
the needs of children arriving with their parents. In this task, policy makers would benefit
from understanding why age at migration predicts lower education. The key challenge in
interpreting our baseline results is that families choosing to migrate with children might
differ in many ways from those whose children are born only after they have settled into
the host country. We examine the importance of such selection by comparing siblings
who grow up in the same families, but arrive in Finland at different ages. This analysis is
likely to capture the causal impact of age at arrival, because siblings share time-invariant
family characteristics that likely lead to selection bias in cross-sectional analysis.
We find that even after conditioning on time-invariant family characteristics, age
at arrival affects educational attainment in an economically and statistically significant
manner. To better understand how this effect arises, we conduct a more detailed analysis.
The results suggest that arriving as a small child has a strong negative effect in comparison
to being born in Finland. This pattern is inconsistent with the age at arrival effects being
primarily driven by the critical period of language acquisition. Furthermore, parents’
exposure to Finland prior to the birth of the child does not seem to have an impact,
suggesting that the length of the child’s own exposure to Finland is the most likely
explanation for why age at arrival matters in our context.
In the final part of the paper, we complement our main analysis by examining factors
2
that might be consequences or causes of low educational attainment. We document
differences in idleness, criminal convictions and the use of medical services. Some of
the results closely mirror the pattern of results for educational attainment. Children of
immigrants are much more likely to be idle (i.e., not to be in education, employment or
training) and to have been sentenced for a crime than are children of natives. Similar to
education, the gaps are driven by those moving to Finland with their parents, whereas
Finnish-born children of natives and immigrants growing up in similar circumstances
seem to have similar outcomes. Interestingly, however, the results for health outcomes
are very different. Children of immigrants are less likely to have a diagnosis of severe
illness or to have purchased psychotropic medication. For psychotropic medication, the
gaps are larger for second-generation immigrants and increase when conditioning on other
background characteristics. These results are somewhat surprising because immigrants
tend to have worse self-reported health than do natives (Matikka et al. 2015). We
interpret this pattern to be most likely due to immigrants using fewer medical services
than do natives with similar underlying health problems.
These results add to a growing literature documenting differences in educational at-
tainment between children of immigrants and children of natives.1 Our contribution to
this literature is twofold. First, we present the first comprehensive documentation of how
children of immigrants cope in the Finnish educational system. Second, and more impor-
tantly, we examine the importance of childhood resources by conditioning on both family
characteristics and childhood neighborhoods. While earlier work has typically shown
that conditioning on parental characteristics—most often parental education—decreases
educational gaps, the role of neighborhoods has received little attention.2 Our results
show that this omission is important. Specifically, while controlling for parental charac-
teristics and family structure reduces the immigrant-native gap in completing secondary
1Earlier work on educational attainment includes Chiswick and DebBurman (2004), Glick and White(2004), and Perreira et al. (2006) for the United States; Dustmann and Theodoropoulos (2010) for theUnited Kingdom; Gang and Zimmermann (2000) and Riphahn (2003) for Germany; Belzil and Poinas(2010) and Dos Santos and Wolff (2011) for France; Baert and Cockx (2013) for Belgium; Van Ours andVeenman (2003) for the Netherlands; Bratsberg et al. (2012) for Norway; Hallsten and Szulkin (2009)for Sweden; and Algan et al. (2010) for France, Germany and the United Kingdom. Earlier work onFinland is limited to that of Kilpi-Jakonen (2011), who examines transitions from middle school to uppersecondary education, whereas we document educational attainment in early adulthood. This distinctionis important in the Finnish context, because differences in educational attainment are largely driven bydropping out from secondary degree programs rather than never enrolling in one.
2Hallsten and Szulkin’s (2009) study seems to be the only earlier study conditioning on relativelysmall neighborhoods.
3
education by a half, adding neighborhood fixed effects to the specification eliminates the
rest of the gap in the Finnish context.
We also contribute to the literature documenting a negative association between age at
arrival and educational attainment.3 Earlier work using sibling comparisons to examine
the role of selection in creating this association include Bratsberg et al. (2012) and
Hermansen (2017) for Norway and Aslund et al. (2015) for Sweden.4 Similar to us, these
studies find that while selection plays a role, age at migration also has a causal impact.
Finally, we add to the research on crime and the health of immigrants’ children.5
Similar to our analysis for educational attainment, our contribution is to present the
first comprehensive overview of these outcomes for immigrant children living in Finland.
Furthermore, our paper is unique in using identical methodology to examine a wide set
of outcomes across several domains. This approach allows us to paint a more nuanced
picture of the situation of immigrants’ offspring and to examine the extent to which
lessons from one set of outcome variables can be extrapolated into other domains.
We proceed as follows. The next section provides a brief introduction to Finland’s
immigration experience and institutional context. We present our data in Section 3, re-
port the results for educational attainment in Section 4, and complement this analysis by
examining crime and the use of medical services in Section 5. We conclude by discussing
potential interpretations of our results and our results’ implications for public policy and
future research.
2 Institutional context
Finland presents an interesting setting to study the integration of the children of immi-
grants because of its record of providing opportunities for children growing up in dis-
advantaged families. Finland has one of the world’s highest rates of intergenerational
3Previous work using cross-sectional data includes Chiswick and DebBurman (2004), Bleakley andChin (2004), Gonzalez (2003), Lee and Edmonston (2011), and Myers et al. (2009) for the United States;Corak (2012), Lee and Edmonston (2011), and Schaafsma and Sweetman (2001) for Canada; Van Oursand Veenman (2006) for the Netherlands; and Sohn (2011) for Germany.
4In addition, Bohlmark (2008) uses sibling comparisons to examine the impact of age at migrationon test scores in Sweden.
5Earlier work on crime includes Bersani (2014), Bui (2009) and Rumbaut et al. (2006) for the UnitedStates; Junger-Tas et al. (2012, 2010) for various European countries; and Kardell and Carlsson (2009)for the Nordic countries. For health outcomes, see surveys by Molcho et al. (2010), Pfarrwaller and Suris(2012) and Stevens and Vollebergh (2008).
4
Figure 1: Children with foreign-born parents living in Finland in 1990-2014
0
10000
20000
30000
40000
50000
60000
70000
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014Year
Born in FinlandBorn abroad
Note: This figure reports the number of under 18-year-old individuals living in Finland with both parents
(or the only known parent) foreign-born. Source: Statistics Finland.
mobility (Corak 2013; Jantti et al. 2006) and the PISA assessments have consistently
shown that Finland has both high average test scores and a weak relationship between
students’ family background and performance (OECD 2013). The Finnish government
has relied on universal policies, that is, virtually no policies are specifically targeted at
children of immigrants.
On the other hand, Finland also has a relatively short immigration history. For most
of the post-WWII period, Finland has been an emigration country, where immigrants
were primarily return migrants and their family members. Finland became a destination
country for immigrants only in the early 1990s. Figure 1 illustrates the pace of this
change by plotting the number of children with foreign-born parents living in Finland
in 1990–2014. Our analysis focuses on the children of immigrants who arrived during a
period when the immigrant population was still small. In 2014, when we measure the
outcomes of our youngest cohorts, 5% of the Finnish population were immigrants and 6%
were categorized as being of a foreign background (i.e., both parents or the only known
parent was born abroad).
The share of economic migrants to Finland has been low. Furthermore, a large share
of the arrival cohorts we examine arrived during an exceptionally severe recession in the
early 1990s. These factors are reflected in the low initial employment rates and incomes
5
of adult immigrants in comparison to those of natives. The gaps decreased over time,
but even after living in Finland for 15 years, immigrants tend to perform substantially
worse in the Finnish labor market than comparable natives (Sarvimaki 2011, 2017).
3 Data, definitions and descriptive statistics
3.1 Data sources and sample restrictions
Population-wide data Our population-wide data cover the Finnish population aged
15–70, who were residents in Finland between 1988 and 2014. Statistics Finland has con-
structed these data by drawing information from several administrative registers, which
contain rich information on individual characteristics, educational attainment, employ-
ment status and income. We also observe identification codes for individuals’ households
and family units, and the zip codes of their places of residence. The sample for which we
examine educational outcomes consists of individuals who were born between 1975 and
1991. We further restrict the sample to those living in Finland continuously between the
ages of 15 and 23. Lastly, we include only individuals who were either born in Finland
or immigrated to Finland at age 15 or younger.
Two factors motivate our cohort restrictions. First, hardly any immigrants’ children
living in Finland were born before 1975. Second, those born in 1991 are the the last
cohort for whom we observe the educational attainment at age 23. We require contin-
uous residence in Finland in adolescent years in order to exclude temporary migrants.
Furthermore, we study immigrants arriving as children or early adolescents in order to
focus on those likely to participate in compulsory education, and to exclude immigrants
arriving as exchange students or for employment purposes.
Crime and health data A limitation of our population-wide data is that they do not
include information on other potentially relevant outcome variables beyond educational
attainment. Thus, we augment our analysis by using another data set consisting of a
20% random sample of the population born between 1967 and 1990. Similar to our
population-wide data, these data are drawn from several administrative registers that
are linked together at Statistics Finland. Importantly, however, this data set includes
detailed information on the purchases of prescription drugs and on criminal sentences. We
6
study individuals’ medication use and criminal sentences by the age of 23, and construct
the sample similarly as described above. However, we observe these outcomes only up
to the end of 2008 and thus limit the sample to individuals born between 1975 and
1985. In other respects, the data set contains background information on individuals
that is similar to, though somewhat less precise than, the background information in the
population-wide data.6
3.2 Definitions of family and immigrant status
We define families in a given year as persons living in the same household and belonging
to the same family unit as defined by Statistics Finland’s family register. ”Parents”
are adults who are first observed in the same household as the individuals we study
(henceforth ”the children”). We consider children to be siblings if they have the same
mother. However, if there is no mother in the same household, we define siblings using
the father’s identification code. We exclude from the sample individuals with no observed
parents.
In the baseline analysis, we group children into three categories according to the origin
of their parents. We call individuals whose parents are both natives ”children of natives,”
those with one immigrant and one native parent ”children of immigrant-natives,” and
those whose parents are both immigrants ”children of immigrants.”7 These definitions do
not depend on whether the child was born in Finland or abroad. If we observe only one
parent, we categorize the child according to the status of that parent (i.e., immigrant or
native). In the analysis with more detailed origin categories, we use the mother’s origin
if we observe her before or at the same time as the child’s father, and use the father’s
origin otherwise.
6In the smaller data set, information on individuals’ place of residence is not as detailed as suchinformation in the population-wide data is. See the Data Appendix for how we construct this variablein the crime and health data.
7We define parents’ immigrant status according to their country of birth and registered language.We define natives as individuals who are either born in Finland or born abroad and whose registerednative language is Finnish or Swedish in the first year we observe them. An exception is Swedish-born,Swedish-speaking individuals, who are defined as immigrants because a majority of them are likely to beSwedes rather than Swedish-born children of Finnish emigrants. Furthermore, we consider individualsborn in the former Soviet Union countries to be immigrants regardless of their registered language. Thoseborn in the Soviet Union but having Finnish as their registered language are likely to be Ingrian Finns,whose standing in Finland is likely to be more comparable to that of immigrants than to that of natives.
7
3.3 Background characteristics
Table 1 presents averages of the background characteristics for the three categories of
children defined in the previous section, measured in the year the children turned 15
years old. It shows that children of immigrants grow up, on average, in households with
low incomes and low parental labor market attachment. They are also more likely to
live in single-parent households. The differences are smaller when we compare native
families to families where one parent is a native and the other is an immigrant. We do
not report the education of the parents (or control for it in our regressions), because
education obtained abroad is poorly measured in our data.8
Importantly, our definition of immigrants’ children includes both those born in Finland
and those who moved to Finland before or at the age of 15. Appendix Table A1 shows
the distribution of age at immigration for the children in our main data. Indeed, 57%
of children of immigrant parents in our sample immigrated between ages 7–15, and only
12% were born in Finland (or arrived as infants). Thus, it is important to bear in mind
that the majority of individuals we examine are not second-generation immigrants.
3.4 Outcomes
Our primary outcome is an indicator for holding an upper secondary degree at age 23.
These degrees are granted by general upper secondary schools (”academic track”) and
vocational upper secondary schools and roughly correspond to high school degrees in the
U.S. context.9 Most 16-year-olds enroll in one of these institutions after completing the
obligatory nine-year comprehensive education. Failing to complete a secondary degree is
a strong predictor of low income, high unemployment, and a poor level of housing (Aro
2009).
Our second measure of educational attainment is an indicator for having ever been
8The Register of Educational Qualifications and Degrees has comprehensive coverage only of degreesattained in Finland. Some degrees of immigrants are recorded in this register via the employment services,but the recording of this information is incomplete and available only for selected groups of immigrants.Using survey data, Sutela and Larja (2015) find that 24% of working-age immigrants had completed atmost lower secondary education, 42% had at most an upper secondary degree, and 35% had a degreefrom higher education. The corresponding shares among persons with a Finnish background were 17%,49% and 35%, respectively.
9We use a relatively old age for measuring secondary education because many children of immigrantsprogress through the educational system with some delay. However in our data, almost all individualswho had completed upper secondary education had graduated by age 23. See the Data Appendix for amore detailed discussion.
8
Table 1: Background characteristics
Immigrant-Native native Immigrant
A: Population-wide sampleParents’ average months of unemployment 1.36 2.05 4.07Parents’ combined income, euros 53,033 48,246 24,278Parents’ combined income, deciles 5.52 4.78 2.37Mother’s age 43.1 43.1 40.9Father’s age 45.2 46.6 43.8Number of under-aged children 1.83 1.83 2.32Age at arrival, defined by child 0.06 2.63 8.55Mother present when child 15 0.97 0.96 0.95Father present when child 15 0.84 0.78 0.74Mother present ever 1.00 1.00 0.99Father present ever 0.97 1.00 0.90
B: Crime and health sampleParents’ average months of unemployment 1.04 1.66 4.39Parents’ combined income, euros 50,465 48,250 20,883Parents’ combined income, deciles 5.63 5.09 2.26Mother’s age 42.7 42.8 40.4Father’s age 44.8 46.0 43.1Number of under-aged children 2.07 2.03 2.38Age at arrival, defined by child 0.07 2.73 10.57Mother present when child 15 0.96 0.95 0.93Father present when child 15 0.85 0.81 0.72Mother present ever 1.00 1.00 0.99Father present ever 0.96 1.00 0.86
Observations (population) 1,039,341 13,045 13,424Observations (crime and health sample) 138,094 1,330 1,039
Note: Average background characteristics are measured in the year the person turns 15 years old.
A person is categorized as a ”native” if both of her parents are natives, as an ”immigrant-native” if
one of her parents is an immigrant and the other is a native, and as an ”immigrant” if both parents
are immigrants. See Section 3.2 and the Data Appendix for variable definitions.
9
Table 2: Outcomes
Immigrant-Native native Immigrant
A: EducationHolds a secondary degree at age 20 0.80 0.69 0.48Holds a secondary degree at age 23 0.86 0.79 0.65Enrolled in higher education by age 23 0.47 0.43 0.29
B: Criminal sentencesSentenced to fines by age 23 0.10 0.13 0.20Sentenced to imprisonment orcommunity service by age 23 0.04 0.05 0.09Sentenced to unconditional imprisonment by age 23 0.01 0.01 0.02
C: Medical servicesPrescribed medication for severe illness 0.04 0.03 0.02Used psychotropic medication 0.11 0.13 0.09
Note: This table reports averages of our outcome variables by parental background. A person is categorized
as a ”native” if both her parents are natives, as an ”immigrant-native” if one of her parents is an immigrant
and the other is a native, and as an ”immigrant” if both her parents are immigrants. See Section 3.2 and the
Data Appendix for variable definitions.
enrolled in higher education between ages 20 and 23.10 In Finland, college education is
provided by universities and universities of applied sciences (also known as polytechnics).
Finns who hold a college degree have substantially higher lifetime income (Koerselman
and Uusitalo 2014), better self-rated health and a lower incidence of long-standing limiting
illness (Laaksonen et al. 2005; Lahelma et al. 2004), and lower mortality and lifespan
variation (Elo et al. 2006; van Raalte et al. 2011) than do Finns holding lower degrees.
Table 2 shows that 80% of the children of natives in our sample held an upper sec-
ondary degree at age 20 and the share increases to 86% by age 23. Among children of
immigrant parents, holding this degree is less common, with less than half the children
of immigrants having attained the degree at age 20, and 63% at age 23. The difference
between children of natives and children of immigrants is apparent also in higher educa-
tion enrollment. While nearly half the children of native parents and native-immigrant
parents had been enrolled in higher education by age 23, the corresponding figure for the
children of immigrants was only 29%.11
10We have information on enrollment beginning from 1995, when the birth cohort of 1975 was age 20.To have the same at-risk period for all cohorts, we measure the outcome variable when the person is ages20 and 23.
11We do not examine employment or earnings, because a large proportion of Finnish youth are stillin education at the age of 23. Thus, income and employment at this age are unlikely to provide goodproxies for their future labor market potential.
10
Panel B of Table 2 reports the share of individuals having been sentenced for a crime
by the District Courts and/or the Courts of Appeal by age 23. A fifth of the children
of immigrants have been sentenced to fines as compared with a tenth of the children of
natives. The relative differences are comparable also for the indicators of more serious
crimes, measured by having been sentenced to conditional imprisonment, community
service or unconditional imprisonment.
The last panel of Table 2 reports two measures of the use of medical services. It shows
that children of immigrants are less likely to be treated for severe and chronic illnesses
or to be prescribed psychotropic medication. We emphasize that these outcomes do not
necessarily measure differences in the underlying health, but rather the combination of
health and the capability (or willingness) to make one’s way through the health care
system. We return to this issue in Section 5.
4 Educational attainment
In the previous section, we documented large differences in the average educational at-
tainment between children of immigrants and children of natives. We now examine the
extent to which these differences reflect the fact that the children of immigrants grow up
in poorer families and in worse neighborhoods. We also examine in detail the role of age
at migration and parental region of origin.
4.1 Baseline results
We start by measuring differences in educational attainment between immigrants’ children
relative to native children in early adulthood by using a linear probability model
yi = α + Iiβ + θAi +Xiγ + εi (1)
where yij is an indicator variable for individual i’s educational attainment, Ii is a vector of
indicator variables for her parents’ immigrant category (using children with both parents
native as the omitted category), Ai is her age at arrival (zero for those born in Finland),
Xi is a vector of year of birth indicators and background characteristics measured at age
15, and εi is an error term. We report estimates for the baseline specification controlling
only for the individual’s year of birth and then gradually add control variables for age at
11
migration, background characteristics and residential location. The standard errors are
clustered at the level of municipality of residence.12
Table 3 reports the results using a specification with the same two categories for
immigrant background as in the previous section. The baseline estimates reported in
column (1) condition only on differences in year of birth and corresponds closely to the
raw differences discussed above. That is, children of families with one native and one
immigrant parent are 10.6 percentage points less likely to have completed secondary
education by age 20 than individuals who grew up in entirely native families. The gap
between all-immigrant and all-native parents is substantially larger at 31.3 percentage
points.
The next column reports estimates from a specification controlling for age at migra-
tion. The estimates reveal large differences across this dimension. Second-generation
immigrants—defined as children of all-immigrant families who are born in Finland—are
12.8 percentage points less likely to hold a secondary degree than children of all-native
families, while the gap is 12.8 + 15 × 2.2 = 45.8 percentage points for those moving to
Finland when they are 15 years old.
In the last two specifications, we add controls for background characteristics and
residential location. Controlling for parental labor market outcomes and family size
reduces the estimates for β by roughly a half among the children of immigrants (column
3, see table notes for details of the control variables). Controlling for the zip code where
the person lived at age 15 further reduces the estimates (column 4). Indeed, we do not
find statistically significant different graduation rates between Finnish-born children of
immigrants and children of natives who grow up in the same zip codes and in similar
families in terms of observable characteristics. However, large differences remain between
those born in Finland and those arriving at an older age.
Panel B of Table 3 reports similar estimates using graduation by age 23 as an outcome
variable. We find a very similar, though somewhat attenuated, pattern as above. These
results suggest that the unconditional differences observed at age 20 largely reflect slower
progress through the educational system. Thus we focus on graduation by age 23 in much
of the further analysis below.
The remaining estimates of Table 3 examine the likelihood of starting higher education
12We have chosen this level of clustering to take into account regional variation rising from, for example,the supply of post-mandatory education and returns to education.
12
Table 3: Educational Attainment in Comparison to that of Children of Natives
(1) (2) (3) (4)
A: Holds a secondary degree at age 20Immigrant-native -0.106 -0.049 -0.034 -0.010
(0.005) (0.005) (0.004) (0.004)
Immigrant -0.313 -0.128 -0.067 -0.008(0.006) (0.007) (0.007) (0.007)
Age at migration . -0.022 -0.015 -0.015(0.001) (0.001) (0.001)
B: Holds a secondary degree at age 23Immigrant-native -0.069 -0.033 -0.020 0.003
(0.004) (0.004) (0.004) (0.004)
Immigrant -0.210 -0.093 -0.041 0.013(0.006) (0.007) (0.006) (0.007)
Age at migration . -0.014 -0.008 -0.008(0.000) (0.000) (0.000)
C: Enrolled in higher education by age 23Immigrant-native -0.034 0.008 0.021 0.035
(0.005) (0.005) (0.005) (0.005)
Immigrant -0.174 -0.037 0.048 0.091(0.006) (0.008) (0.007) (0.007)
Age at migration . -0.016 -0.006 -0.006(0.001) (0.001) (0.001)
Controlling for:Parental characteristics no no yes yesResidence location ZIP-codes no no no yes
Note: This table reports estimates from regressing indicators of educational attain-
ment on indicators of immigrant background. A person is categorized as a ”native”
if both of her parents are natives (the omitted category), as an ”immigrant-native”
if one of her parents is an immigrant and another one a native, and as an ”im-
migrant” if both parents are immigrants. Columns 2–4 also condition on age at
migration, column 3 adds controls for background characteristics and column 4
for residential location. Background characteristics are parents’ average months of
unemployment (14 categories), income decile of parents’ combined income (ten cat-
egories), mother’s and father’s age (seven categories) and number of siblings (six
categories). Residential location is measured as the zip code of residence. All back-
ground characteristics are measured at age 15. The outcomes are an indicator for
holding a secondary degree at age 20 (panel A) or at age 23 (panel B) and having
been enrolled in a university or a polytechnic by age 23 (panel C).
13
by age 23. Again, we find a large unconditional difference by parents’ immigrant status,
which is much larger for children who arrived to Finland at later ages. However, second-
generation immigrants are 9.1 percentage points more likely to study in a university or
polytechnic than children of natives with similar background characteristics.
4.2 Heterogeneity by region of origin
We next turn to differences in educational attainment across parents’ region of origin.
Figure 2 reports estimates for region of origin fixed-effects from an extended version of
equation (1), where vector Ii now includes separate indicators for 11 country/region of
origin fixed-effects. We report both unconditional estimates (corresponding to the first
colum of Table 3) measuring the overall differences between children of immigrants and
natives, and conditional estimates (corresponding to the last column of Table 3).
The unconditional estimates reveal substantial heterogeneity by region of origin. For
the likelihood of graduating from an upper secondary school (top panel), the uncondi-
tional gaps vary between 39 and 52 percentage points among children of immigrants from
Somalia, Turkey and Iraq, while the differences are much smaller for the children of im-
migrants from European countries. The conditional gaps are substantially smaller. In
fact, the only statistically significant negative estimate is for the children of immigrants
from Somalia. These results suggest that differences in educational attainment between
region of origin groups can be largely attributed to their differences in age at arrival,
parental characteristics and neighborhoods.
The baseline results for attending college (bottom panel) are qualitatively similar.
Interestingly, however, the estimates suggest that Finnish-born children of immigrants
from most regions of origin are more likely to have attended higher education by age 23
than are children of natives growing up in similar circumstances. This is the case even
among the children of Somalian immigrants, that is, the only group for which we found
a statistically negative estimate for secondary education.
4.3 Sibling comparisons
Above we saw that children of immigrants arriving in Finland at a later age have lower
educational attainment than do those born in Finland. Interpreting this association is
challenging, because families choosing to migrate with older children might differ from
14
Figure 2: Immigrant-Native Gaps in Educational Attainment
Europe, USA, Can, AUS, NZ
form. Soviet Union
form. Yugoslavia
Asia (other)
Latin America
Vietnam
Iran
Africa (other)
Iraq
Turkey
Somalia
-0.07
-0.13
-0.21
-0.21
-0.21
-0.22
-0.24
-0.29
-0.39
-0.47
-0.52
0.04
0.01
-0.01
0.05
-0.11
0.04
-0.04
-0.05
-0.04
0.00
-0.13
-0.50 -0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30 0.40
BaselineFull controls95% CI
Europe, USA, Can, AUS, NZ
form. Soviet Union
Asia (other)
Vietnam
Iran
Africa (other)
form. Yugoslavia
Turkey
Iraq
Latin America
Somalia
-0.00
-0.11
-0.14
-0.18
-0.22
-0.23
-0.28
-0.32
-0.33
-0.34
-0.35
0.05
0.06
0.18
0.13
0.07
0.13
-0.02
0.03
0.02
-0.23
0.13
-0.50 -0.40 -0.30 -0.20 -0.10 0.00 0.10 0.20 0.30 0.40
B. Enrolled in higher education by age 23
Note: This figure reports region of origin fixed-effects from regressions corresponding to specifications
(1) and (5) of Table 3, panel B, but replacing the immigrant indicator with region of origin indicators.
15
Table 4: Cross-Sectional and Within-Family Estimates for Age at Migration
NativeImmigrants Immigrant-Natives return migrants
(1) (2) (3) (4) (5) (6)
Secondary degree -0.029 -0.015 -0.027 -0.015 -0.019 0.001at age 20 (0.004) (0.006) (0.004) (0.003) (0.000) (0.000)
Secondary degree -0.014 -0.010 -0.018 -0.010 -0.015 0.002at age 23 (0.003) (0.004) (0.003) (0.002) (0.000) (0.000)
Enrolled in higher -0.015 -0.004 -0.022 -0.004 -0.019 -0.002education by age 23 (0.003) (0.002) (0.003) (0.002) (0.000) (0.000)Family FEs no yes no yes no yes
Note: Point estimates and robust standard errors (in parentheses) from regressing indicators for
educational outcomes on age at arrival and year indicators. Estimates in columns (2), (4) and (6)
control for family fixed-effects, gender and being the first-born child. Each estimate stems from a
separate regression. Age at arrival is set to zero for persons born in Finland.
other families along dimensions that also affect children’s educational attainment. In
this section, we present comparisons between siblings who grow up in the same families,
but arrive in Finland at different ages. Because siblings share time-invariant family-level
characteristics, these estimates are less likely to suffer from selection bias and thus are
more likely to capture the causal impact of age at migration than are the cross-sectional
estimates discussed above.
Table 3 shows that the cross-sectional estimates are large and negative among immi-
grant and immigrant-native families. The within-family estimates are somewhat smaller
than the cross-sectional estimates, but remain economically and statistically significant.
For example, the estimates suggest that children of immigrants arriving at age 15 are 15
percentage points less likely to hold an upper secondary degree, and 6 percentage points
less likely to have attended college by age 23 than their Finnish-born siblings are.
In the last columns of Table 3, we report similar estimates for children of native
emigrants returning to Finland. The estimates show a strong negative association between
age at arrival and educational attainment also for this group. However, conditioning on
family fixed-effects yields precisely estimated zeros. Thus the negative association in the
cross-section appears to be driven purely by selection among native return migrants.
Figure 3 presents non-parametric versions of the within-family estimates for holding
an upper secondary degree at age 23. In this figure, we also examine the role of parents’
time in Finland prior to the birth of the child and thus define age at arrival using parents’
16
Figure 3: Within-Family Estimates for the Effect of Age at Migration on Holding anUpper Secondary Degree at Age 23
-.4
-.3
-.2
-.1
0
.1
.2
.3
-5 or more -4-3 -2-1 0-1 2-3 4-5 6-7 8-9 10-11 12-13 14-15Age at arrival
Native return migrantsImmigrant-nativesImmigrants
Note: Within-family estimates and 95% confidence intervals for age at arrival.
migration year.13 We find no evidence that parents’ exposure to Finland prior to the birth
of the child would affect the child’s later educational attainment. Interestingly, among
the children of immigrants, there seems to be a large negative impact of arriving at ages
2–3 (in comparison to being born in Finland), but we find no additional impact between
ages 4–7. The point estimates suggest that age at arrival starts to matter more after
age 8, but the estimates are too imprecise for drawing strong conclusions. Importantly,
given the imprecision of the estimates, we also cannot rule out a linear effect throughout
arrival ages. The point estimates for the children of immigrant-natives suggest a roughly
linear impact among native-immigrant families up to age 13 and a larger drop for those
arriving at ages 14–15. Again, we find little evidence that age at migration would affect
educational attainment of the children of native return migrants.
These results suggest that age at migration has a causal impact on the educational
attainment of children growing up in immigrant and immigrant-native families. This
effect could arise for many reasons and understanding its origins would be important for
designing appropriate policy responses. While we do not have a strong research design
13For example, an age of arrival of −1 means that the child was born a year after the parent arrivedin Finland.
17
for examining potential mechanisms, we note that the pattern documented above is more
consistent with some explanations than with others. For example, it seems unlikely that
migration related disruption in social networks and education would be a primary reason
behind age at migration effects, because such effects would likely be present also among
the children of native return migrants.
One reason for why age at migration may matter is that children arriving after the
end of the critical period of language acquisition—typically timed before puberty—are
less likely to become fluent in the local language (see Bleakley and Chin (2004, 2010),
and references therein). Language deficits could then make school harder for children
arriving at later ages and lower their educational attainment. However, this mechanism
would yield a pattern, where children arriving in early and middle childhood would cope
as well as those born in Finland, while we would see a sharp drop in performance for
those arriving as teenagers. In contrast, the clearest pattern present in Figure 3 is a large
drop between children of immigrants arriving at age 2–3 in comparison with those born
in Finland. Furthermore, we find strong age at arrival effects also for those who have one
immigrant and one native parent, even though they are more likely to learn at least some
Finnish or Swedish from their native parent.14
Another potential mechanism is that age at arrival could capture the importance of
parents’ exposure to Finland. A large body of literature has shown that integration of
adult immigrants into the host country’s labor market and broader society tends to take
time. Thus, children arriving with their parents spend a larger share of their childhood
in the period when their parents are more likely to struggle in the labor market than
children born after their parents have already settled in Finland. However, if parents’
exposure to Finland would be the primary mechanism behind age at migration effects,
an additional year in Finland before the birth of the child should have a roughly similar
impact as an additional year after the child is born. In contrast, Figure 3 suggests that
this is not the case in our data.
Finally, longer exposure to Finnish institutions could improve educational attainment.
The importance of this effect is likely to depend on the quality of institutions in the source
countries. For example, children of natives living abroad are likely to live in countries
resembling Finland in many ways, while children of immigrants may grow up in very
14Finland is a bilingual country, where roughly 5% of the population speak Swedish as their mothertongue.
18
different circumstances before arriving to Finland.15 The patterns presented in Figure 3
are consistent with the institutional explanation. Based on point estimates, one could
infer that institutions affecting children of immigrants in early childhood are particularly
important. However, given the low precision of our estimates we do not believe one
should draw strong conclusions from our results on the relative importance of exposure
to Finland at different ages.
5 Other outcomes
We end our analysis with a brief look at being idle (NEET, i.e., not in employment,
education or training), criminal sentences and the use of medical services. The motivation
for this analysis is to provide a more nuanced picture of the children of immigrants in
their early adulthood. A limitation of this analysis is that the data on crime and the
use of medical services are less detailed and cover a shorter period than our data on
educational attainment do (see Section 3 for details). Nevertheless, we have used as
comparable an approach as possible to that for educational attainment. For comparison,
Appendix Table A2 reproduces our analysis for educational attainment using these more
limited data.
5.1 Idleness
The first panel of Table 5 reports the estimates for an indicator for being outside of
employment, education and training at age 23. The structure is similar to that of Table
3, that is, the first column reports the baseline differences controlling only for year of
birth indicators and the subsequent columns report estimates from specifications where
we gradually add controls for age at arrival, background charateristics and neighborhood
fixed-effects. The results closely mirror those for educational attainment. Again, we
find a large unconditional gap between children of immigrants and natives, which is
largely driven by those immigrating to Finland at older ages. Once we condition on other
background characteristics, we find that second-generation immigrants are less likely to be
idle at age 23 than children of natives growing up in comparable circumstances. However,
15The largest destination countries for Finnish emigrants in 2017 were Sweden, the United Kingdom,the United States, Germany and Spain.
19
Table 5: Idleness and Criminal Sentences by Age 23 in Comparison to the Children ofNatives
(1) (2) (3) (4)
A: Not in employment, education or training at age 23Immigrant-native 0.028 0.004 -0.004 -0.003
(0.003) (0.003) (0.003) (0.003)
Immigrant 0.099 0.022 -0.016 -0.020(0.004) (0.006) (0.006) (0.006)
Age at migration . 0.009 0.005 0.005(0.000) (0.000) (0.000)
B: Sentenced to fines by age 23Immigrant-native 0.036 0.020 0.020 0.013
(0.009) (0.008) (0.007) (0.008)
Immigrant 0.097 0.035 0.017 0.008(0.009) (0.019) (0.016) (0.015)
Age at migration . 0.006 0.002 0.002(0.001) (0.001) (0.001)
C: Sentenced to imprisonment or community service by age 23Immigrant-native 0.011 -0.001 -0.002 -0.005
(0.006) (0.006) (0.006) (0.006)
Immigrant 0.052 0.006 -0.007 -0.012(0.013) (0.015) (0.013) (0.013)
Age at migration . 0.004 0.002 0.002(0.001) (0.001) (0.001)
D: Sentenced to unconditional imprisonment by age 23Immigrant-native 0.001 -0.001 -0.001 -0.003
(0.003) (0.003) (0.003) (0.003)
Immigrant 0.012 0.006 0.002 -0.001(0.003) (0.004) (0.004) (0.004)
Age at migration . 0.001 0.000 0.000(0.000) (0.000) (0.000)
Controlling for:Parental characteristics no no yes yesResidence location no no no yes
Note: This table reports estimates from regressing indicators of NEET status(panel A) and criminal convictions (panels B–D) on indicators of immigrantbackground. A person is categorized as a ”native” if both of her parentsare natives (the omitted category), as an ”immigrant-native” if one of herparents is an immigrant and another one a native, and as an ”immigrant” ifboth parents are immigrants. Columns 2–4 also condition on age at migra-tion, column 3 adds controls for background characteristics and column 4 forresidential location. In Panel A, residential location is measured as the zipcode of residence. In the remaining panels, residential location is measuredas the interaction of province of residence and the level of urbanization of theresidence municipality. All background characteristics are measured at age15. See the note to Table 3 and the Data Appendix for variable definitions.
20
the gap remains large for those immigrating to Finland at age 15.
5.2 Criminal sentences
The remainder of Table 5 reports the results for criminal sentences. Similar to educational
attainment, we document a statistically and economically significant difference between
immigrants’ and natives’ children. Children of immigrants are 10 percentage points more
likely to have been sentenced to fines than are children of natives. These sentences are
given predominantly due to traffic violations such as aggravated endangering of traffic
safety or drunk driving. We also find large differences in the likelihood of being sentenced
for more serious crimes. The difference is 5 percentage points for having been sentenced
to imprisonment or community service, and 1 percentage point for having served prison
time. In relative terms, children of immigrants are roughly twice as likely to have received
a conviction than are the children of natives.
Again, however, the differences are largely driven by those arriving to Finland at a
later age.16 As shown in column 2 of Table 5, we find no difference between second-
generation immigrants and children of natives (nor those with one immigrant and one
native parent) in the likelihood of having been sentenced to community service or to
imprisonment. A significant difference remains for being sentenced to fines, but also this
difference disappears once we condition on other background characteristics.
To interpret the results correctly, it is important to note that the differences in sentence
prevalence might capture differences in both the underlying crime rates and/or differen-
tial treatment within the justice system. In their overview, for example, Junger-Tas and
Marshall (1999) point out that for some immigrant groups, studies of self-reported crime
often do not display equally large differences in the prevalence of crime between immi-
grants and natives as conviction rates do. On the other hand, the validity of self-reported
data might also vary across ethnic groups (e.g., van Batenburg-Eddes et al. 2012).
One way to assess this issue in our context is to contrast our findings to earlier work on
self-reported crime. Salmi et al. (2015) find that teenagers of immigrant background self-
report more delinquency, violent behaviour and, in particular, drug use than teenagers
of native origin in Finland. While we cannot directly compare our results to this earlier
16Unfortunately, as we have only a 20% sample at our disposal, we cannot execute a detailed analysisfor the age at migration for these outcomes. The small sample size also limits our possibility to examineheterogeneity by country of origin.
21
work, their results suggest that the differences in criminal sentences are not, at least
entirely, driven by differential treatment for children of immigrants and natives in the
Finnish justice system.
5.3 Medical services
Table 6 reports the results for our measures of the use of medical services by age 23.
Panel A shows that children of immigrants are 1.5 percentage points less likely to have
a diagnosis of severe illness than native children do. This is a large difference given that
4% of the children of natives have been prescribed medication for a severe illness. Unlike
for our other outcomes, the immigrant-native gap remains stable when we condition for
age at migration and other background characteristics.
Panel B of Table 6 shows that children of immigrants are also less likely to have
been reimbursed for purchasing psychotropic medication than children of natives. Again,
the difference is large at 4.4 percentage points in comparison to 11% baseline among
children of natives. Those arriving at later ages seem to be more likely to be treated
for psychological problems than children of immigrants born in Finland, although the
estimate for age at migration loses statistical significance once we condition on other
background characteristics. Furthermore, the gap between second-generation immigrants
and natives increases when we control for other background variables. The estimate from
the full specification suggests that native children are more than twice as likely to have
been reimbursed a drug primarily designed for mental health problems than second-
generation immigrants growing up in similar circumstances.
It is important to note that these results do not necessarily measure differences in
underlying health, but rather the combination of health and the capability (or willing-
ness) to make one’s way through the health care system. Indeed, in large-scale surveys
conducted in Finnish middle-schools, children of immigrants report more adverse health
symptoms (e.g. fatigue, headaches and anxiety) and difficulties in getting access to a
school nurse or social worker than children of natives do (Matikka et al. 2015).17 Never-
theless, immigrants are less likely to use primary health care services or to visit a hospital
than natives of the same age and gender (Gissler et al. 2006). Thus, a reasonable working
17Castaneda et al. (2012) also report that adult immigrants from selected source countries demonstratemore symptoms of depression and anxiety than do members of the native population.
22
Table 6: Use of Medical Services by Age 23 in Comparison to that of Children of Natives
(1) (2) (3) (4)
A: Prescribed medication for severe illnessNative-immigrant -0.006 -0.005 -0.005 -0.006
(0.006) (0.005) (0.005) (0.005)
Immigrant -0.015 -0.011 -0.013 -0.014(0.004) (0.007) (0.007) (0.007)
Age at migration . 0.000 -0.001 -0.001(0.001) (0.001) (0.001)
B: Used psychotropic medicationNative-immigrant 0.009 0.002 -0.001 -0.009
(0.009) (0.008) (0.007) (0.008)
Immigrant -0.044 -0.068 -0.076 -0.094(0.009) (0.013) (0.013) (0.013)
Age at migration . 0.002 0.001 0.001(0.001) (0.001) (0.001)
Controlling for:Year of birth yes yes yes yesParental characteristics no no yes yesResidential location no no no yes
Note: This table reports estimates from regressing indicators of the use
of medical services on immigrant background. A person is categorized as
a ”native” if both of her parents are natives (the omitted category), as
”immigrant-native” if one of her parents is an immigrant and another one a
native, and as an ”immigrant” if both parents are immigrants. Columns 2–4
also condition on age at migration, column 3 adds controls for background
characteristics and column 4 for residential location. Residential location
is measured as the interaction of province of residence and the level of
urbanization of residence municipality. All background characteristics are
measured at age 15. See note to Table 3 and the Data Appendix for variable
definitions.
23
hypothesis seems to be that immigrants use fewer medical services than do natives with
similar underlying health problems.
6 Conclusions
We examined children of immigrants in early adulthood in Finland with the aim to
document a set of policy-relevant facts that also provide a basis for future research. Our
starting point was the strikingly low educational attainment of immigrants’ children. Less
than half of them hold an upper secondary degree at age 23 and are thus at a high risk of
struggling in the Finnish labor market. Therefore, policies that increase their educational
attainment would likely yield high returns to the public investment. While a descriptive
analysis like ours cannot directly show which policies would be most efficient, our results
provide insight into where policy experimentation and further research could be most
fruitful.
Three results deserve particular attention. First, our estimates suggest that Finnish-
born children of immigrants obtain at least as much education as children of natives grow-
ing up in similar circumstances in terms of family income, family structure and residential
neighborhood. This finding is consistent with the hypothesis that the low educational
attainment of second-generation immigrants largely reflects lower childhood resources
rather than immigrant background in itself. In this case, general policies targeted to
all disadvantaged children—as well as policies improving the labor market performance
of first-generation immigrants—may be an efficient way to help also second-generation
immigrants. We believe that examining this hypothesis more closely using appropriate
research designs is an important topic for future research.
Second, we document large heterogeneity in educational attainment across immigrant
groups. Differences between region of origin can be largely attributed to differences in
family resources and neighborhoods, but sibling comparisons suggest that age at mi-
gration has a strong causal impact on educational attainment. Thus those arriving as
children or adolescents could benefit from interventions that target their specific needs.
Rigorous testing of alternative interventions would be relatively easy to implement and
thus a valuable direction for policy experimentation.
Finally, we show that while the results for idleness and criminal convictions are very
24
similar to those for educational attainment, the patterns are very different for the use
of medical services. Given that immigrants and their children have lower self-reported
health than children of natives in Finland, the fact that they use less psychotropic med-
ication and have fewer diagnoses for a severe illness is likely to arise either from lower
institutional knowledge or from lower willingness to seek help through formal medical ser-
vices. Importantly—and in contrast to the other outcomes we examine—the differences
remain stable or even increase when we compare Finnish-born children of immigrants
with natives’ children growing up in similar circumstances. Thus, addressing the poten-
tial underuse of medical services by immigrants’ children might require different kinds of
policy interventions than those targeting education and crime.
25
References
Algan, Y., Dustmann, C., Glitz, A., Manning, A. (2010) The Economic Situation of
First and Second-Generation Immigrants in France, Germany and the United Kingdom.
Economic Journal 120 (542): F4–F30.
Aro, M. (2009) Koulutus ja Hyvinvointipuutteet (Education and Deficiencies in Wel-
fare). Yhteiskuntapolitiikka 74 (5): 497–508.
Aslund, O., Bohlmark, A., Skans, O.N. (2015) Childhood and Family Experiences and
the Social Integration of Young Migrants. Labour Economics 35: 135–144.
Baert, S., Cockx, B. (2013) Pure Ethnic Gaps in Educational Attainment and School
to Work Transitions: When Do They Arise? Discussion paper No. 7262. IZA Institute
of Labor Economics.
Belzil, C., Poinas, F. (2010) Education and Early Career Outcomes of Second-
Generation Immigrants in France. Labour Economics 17: 101–110.
Bersani, B. (2014) An Examination of First and Second-Generation Immigrant Offend-
ing Trajectories. Justice Quarterly 31 (2): 315–343.
Bleakley, H., Chin, A. (2010) Age at Arrival, English Proficiency, and Social Assimilation
Among US Immigrants. American Economic Journal: Applied Economics 2 (1): 165–
192.
Bleakley, H., Chin, A. (2004) Language Skills and Earnings: Evidence from Childhood
Immigrants. The Review of Economics and Statistics 86 (2): 481–496.
Bratsberg, B., Raaum, O., Røed, K. (2012) Educating Children of Immigrants: Closing
the Gap in Norwegian Schools. Nordic Economic Policy Review. Economics of Educa-
tion, Copenhagen: Nordis Ministerrad, pp. 211–251.
Bohlmark, A. (2008) Age at Immigration and School Performance: A Siblings Analysis
Using Swedish Register Data. Labour Economics 15: 1366–1387.
Bui, H.N. (2009) Parent-Child Conflicts, School Troubles, and Differences in Delin-
quency across Immigration Generations. Crime & Delinquency 55 (3): 412–441.
26
Castaneda, A.E., Rask, S., Koponen, P., Molsa, M., Koskinen, S. (eds.). (2012) Maa-
hanmuuttajien Terveys ja Hyvinvointi. Tutkimus Venalais-, Somalialais- ja Kurditaus-
taisista Suomessa (Health and Well-Being of Immigrants. Research on Immigrants with
Russian, Somali and Kurdish Background in Finland). Terveyden ja hyvinvoinnin laitos
Raportti 61/2012. Tampere: Juvenes Print, Suomen Yliopistopaino Oy.
Chiswick, B., DebBurman, N. (2004) Educational Attainment: Analysis by Immigrant
Generation. Economics of Education Review 23 (4): 361–379.
Corak, M. (2013) Income Inequality, Equality of Opportunity, and Intergenerational
Mobility. Discussion Paper No. 7520. IZA Institute of Labor Economics.
Corak, M. (2012) Age at Immigration and the Education Outcomes of Children. Dis-
cussion Paper No. 6072. IZA Institute of Labor Economics.
Dos Santos, M.D., Wolff, F.-C. (2011) Human Capital Background and the Educational
Attainment of Second-Generation Immigrants in France. Economics of Education Review
30 (5): 1085–1096.
Dustmann, C., Frattini, T., Lanzara, G. 2012. Educational Achievement of Second-
Generation Immigrants: An International Comparison. Economic Policy 27 (69): 143–
165.
Dustmann, C. Theodoropoulos, N. (2010) Ethnic Minority Immigrants and their Chil-
dren in Britain. Oxford Economic Papers, New Series 62 (2): 209–233.
Elo, I.T., Martikainen, P., Smith, K.P. (2006) Socioeconomic Differential in Mortality
in Finland and United States: The Role of Education and Income. European Journal of
Population 22 (2): 179–203.
Gang, I.N., Zimmermann, K.F. (2000) Is Child like Parent? Educational Attainment
and Ethnic Origin. The Journal of Human Resources 35 (3): 550–569.
Gissler, M., Malin, M., Matveinen, P., Sarvimaki, M., Kangasharju, A. (2006) Maahan-
muuttajat ja Julkiset Palvelut (Immigrants and Public Services). Tyopoliittinen tutkimus
296. Helsinki: Tyo– ja elinkeinoministerio.
27
Glick, J.E., White, M.J. (2004) Post-Secondary School Participation of Immigrant and
Native Youth: The Role of Familial Resources and Educational Expectations. Social
Science Research 33 (2): 272–299.
Gonzalez, A. (2003) The Education and Wages of Immigrant Children: The Impact of
Age at Arrival. Economics of Education Review 22 (2): 203–212.
Hallsten, M., Szulkin, R. (2009) Families, Neighborhoods, and the Future: The Tran-
sition to Adulthood of Children of Native and Immigrant Origin in Sweden. Working
paper 2009:9. The Stockholm University Linnaeus Center for Integration Studies.
Hermansen, A.S. (2017) Age at Arrival and Life Chances Among Childhood Immigrants.
Demography 54 (1): 201–229.
Junger-Tas, J., Marshall, I.H., Enzmann, D., Killias, M., Steketee, M., Gruszczynska,
B. (eds.). (2012) The Many Faces of Youth Crime. Contrasting Theoretical Perspectives
on Juvenile Delinquency across Countries and Cultures. New York: Springer.
Junger-Tas, J., Marshall, I.H., Enzmann, D., Killias, M., Steketee, M., Gruszczynska,
B. (eds.). (2010) Juvenile Delinquency in Europe and Beyond. Results of the Second
International Self-Report Delinquency Study. New York: Springer.
Junger-Tas, J., Marshall, I.H. (1999) The Self-Report Methodology in Crime Research.
Crime and Justice 25: 291–367.
Jantti, M., Røed, K., Naylor, R., Bjorklund, A., Bratsberg, B., Raaum, O., Osterbacka,
E., Eriksson, T. (2006) American Exceptionalism in a New Light: A Comparison of
Intergenerational Earnings Mobility in the Nordic countries, the United Kingdom and
the United States. Discussion Paper No. 1938. IZA Institute of Labor Economics.
Kardell, J., Carlsson, K.-M. (2009) Lagforingar av Invandrare och Invandrares Barn i de
Nordiska Landerna (Prosecution of Immigrants and Immigrant Children in the Nordic
countries). Nordisk Tidskrift for Kriminalvitenskab 96: 237–261.
Kilpi-Jakonen, E. (2011) Continuation to Upper Secondary Education in Finland: Chil-
dren of Immigrants and the Majority Compared. Acta Sociologica 54 (1): 77–106.
28
Koerselman, K., Uusitalo, R. (2014) The Risk and Return of Human Capital Invest-
ments. Labour Economics 30: 154–163.
Laaksonen, M., Rahkonen, O., Martikainen, P., Lahelma, E. (2005) Socioeconomic Po-
sition and Self-Rated Health: The Contribution of Childhood Socioeconomic Circum-
stances, Adult Socioeconomic Status, and Material Resources. American Journal of
Public Health 95 (8): 1403–1409.
Lahelma, E., Martikainen, P., Laaksonen, M., Aittomaki, A. (2004) Pathways between
Socioeconomic Determinants of Health. Journal of Epidemiology & Community Health
58 (4): 327–332.
Lee, S.M., Edmonston, B. (2011) Age-at-Arrival’s Effects on Asian Immigrants’ So-
cioeconomic Outcomes in Canada and the U.S. International Migration Review 45 (3):
527–561.
Marttunen, M. (2006) The Functioning of the Juvenile Criminal Justice System in Fin-
land, 1990–2005. In P. Honkatukia, J. Kivivuori (eds.), Juvenile Crime in Finland:
Trends, Causes and Control, National Research Institute of Legal Policy Publications
221, Helsinki: National Research Institute of Legal Policy, pp. 379–81.
Matikka, A., Luopa, P., Kivimaki, H., Jokela, J., Paananen, R. (2015) The Well–Being
of Eighth-Year and Ninth-Year Pupils with a Migrant Background. School Health Pro-
motion Study 2013. Discussion Paper 6/2015. National Institute for Health and Welfare.
Molcho, M., Cristini, F., Gabhainn, S.N., Santinello, M., Moreno, M., Gaspar de Matos,
M., Bjarnason, T., Baldassari, D., Due, P. (2010) Health and Well-being among Child
Immigrants in Europe. Eurohealth 16 (1): 20–23.
Myers, D., Gao, X., Emeka, A. (2009) The Gradient of Immigrant Age-at-Arrival Effects
on Socioeconomic Outcomes in the U.S. International Migration Review 43 (1): 205–229.
OECD. (2013) PISA 2012 Results in Focus. What 15-Year-Olds Know and What They
Can Do with What They Know. Paris: OECD.
Perreira, K.M., Harris, K.M., Lee, D. (2006) Making it in America: High School Com-
pletion by Immigrant and Native Youth. Demography 43 (3): 511–536.
29
Pfarrwaller, E. Suris, J.-C. (2012) Determinants of Health in Recently Arrived Young
Migrants and Refugees: A Review of the Literature. Italian Journal of Public Health 9
(3): 1–16.
Riphahn, R.T. (2003) Cohort Effects in the Educational Attainment of Second Gen-
eration Immigrants in Germany: An Analysis of Census Data. Jounral of Population
Economics 16 (4): 711–737.
Rumbaut, R.G., Gonzales, R.G., Komaie, G., Morgan, C.V., Tafoya-Estrada, R. (2006)
Immigration and Incarceration: Patterns and Predictors of Imprisonment among First-
and Second-Generation Young Adults. In R. Martınez Jr, A. Valenzuela Jr. (eds.), Im-
migration and Crime – Ethnicity, Race and Violence, New York: New York University
Press, pp. 64–89.
Salmi, V., Kivivuori, J., Aaltonen, M. (2015) Correlates of Immigrant Youth Crime in
Finland. European Journal of Criminology 12 (6): 681–699.
Sarvimaki, M. (2017) Labor Market Integration of Refugees in Finland. In T.M. Ander-
sen, A. Piil Damm, O. Aslund (eds.), Nordic Economic Policy Review. Labour Market
Integration in the Nordic Countries, Copenhagen: Nordis Ministerrad, pp. 91–114.
Sarvimaki, M. (2011) Assimilation to a Welfare State: Labor Market Performance and
Use of Social Benefits by Immigrants to Finland. The Scandinavian Journal of Eco-
nomics 13 (3): 665–688.
Schaafsma, J., Sweetman, A. (2001) Immigrant Earnings: Age at Immigration Matters.
The Canadian Journal of Economics 34 (4): 1066–1099.
Schnepf, S.V. (2007) Immigrants’ Educational Disadvantage: An Examination across
Ten Countries and Three Surveys. Journal of Population Economics 20 (3): 527–545.
Sohn, J. (2011) Immigrants’ Educational Attainment: A Closer Look at the Age-at-
Migration Effect. In M. Wingens, M. Windzio, H. de Valk, C. Aybek (eds.), A Life-
Course Perspective on Migration and Integration, New York: Springer, pp. 27–54.
Stevens, G.W.J.M., Vollebergh, W.A.M. (2008) Mental Health in Migrant Children.
Journal of Child Psychology and Psychiatry, 43 (3): 276–294.
30
Sutela, H., Larja, L. (2015) Koulutusrakenne (Educational Composition). In T. Niem-
inen, H. Sutela, U. Hannula (eds.), Ulkomaista syntyperaa olevien tyo ja hyvinvointi
Suomessa 2014 (The Work and Wellbeing of Persons with a Foreign Background in
Finland 2014), Helsinki: Statistics Finland, pp. 29–42.
Van Batenburg-Eddes, T., Butte, D., van de Looij-Jansen, P. Schiethart, W., Raat, H.,
de Waart, F., Jansen, W. (2012) Measuring Juvenile Delinquency: How do Self-Reports
Compare with Official Police Statistics? European Journal of Criminology 9 (1): 23–37.
Van Ours, J.C., Veenman, J. (2006) Age at Immigration and Educational Attainment
of Young Immigrants. Economics Letters 90 (3): 310–316.
Van Ours, J.C., Veenman, J. (2003) The Educational Attainment of Second-Generation
Immigrants in the Netherlands. Journal of Population Economics 16 (4): 739–753.
Van Raalte, A.A., Kunst, A., Deboosere, P., Leinsalu, M., Lundberg, O., Martikainen,
P., Strand, B.H., Artnik, B., Wojtyniak, B., Mackenbach, J.P. (2011) More Variation in
Lifespan in Lower Educated Groups: Evidence from 10 European Countries. Interna-
tional Journal of Epidemiology 40 (6): 1703–1714.
31